Research On Grid Resource Discovery Based On User Behavior

Grid is an advanced network infrastructure in information techniques and provides users with a virtual, unified, transparent computing environment, which purpose is to integrate distributed resources in Internet into a whole system effectively such as computing resources, storage resources, communication resources, information resources et al.. And as an advanced, efficient and intelligent network environment, grid should enable users to fetch, publish, share and manage resources conveniently. And it could help users in Internet to acquire ability in knowledge innovation, collaboration work, problem solving, decision making and even supporting resource management and discovery in higher level. Grid can not only integrate resources and schedule tasks in network to make complex problems sovled with high computing power, but also provide plentiful personalized service to server users with accurate services under the virtual environment. As we all know, the main objects in grid are grid users and resources, and grid mainly need provide users with resources sharing. So we can say that one of the most basic functions of grid is to let users share all kinds of resources in a open network reasonably. Grid resource discovery is one of the important research issues in grid resource management, which is key part to connect resources with resource requesters. It is very important to realize rapid resource location accurately and task scheduling on demand. With development of grid, the requests of users in grid are changing ceaselessly with evolution of type and quantity of grid resource. At the same time, the grid resource discovery should develop with the evolution in grid. However, the present resource discovery methods come from resource management and discovery in traditional distributed computing system, and they can not meet the need on resource matching, resource retrieval, resouece organization and some other key techniques in grid. Those present methods are limited on efficiency, scalability, adaptability and can not support grid evolving dynamically. There are some very important issues to adapt to the development of the grid and to solve the new problems in grid resource discovery. If we can handle these issues well, the grid will develop into a virtual society to supply people with better services.Considering grid framework and grid resources are developing and evoluting, the grid becomes more and more complex. While the relationship between grid users, as resource consumers, is more and more complicated in the open interaction system, researchers must propose some effective way to locate and organize resource which should discover more resource information through analyzing resource description. So that, grid can supply users with accurate services, and it is the future research directions on grid. In practice, to attract resource owners join in collaboration work, grid should protect the resources and owners and maintain the good performance of the system. In the light of the new characteristics in grid, this paper analyzes grid user behavior, interaction mode from interaction information in grid. So that, there are several views on grid resource discovery based on user behavior.(1) Since users'relationship is complicated, users could be consumers and providers on information resources and services in an open interactive system. So researcher should pay more attention on how to organize and exchange resources rationally between users under the complex systems and the problem is future research on grid. This paper analyzes resources exchanging model and their relationship from the view of social interaction theory. Then, the paper puts forward credit cooperation mechanism in grid self-organizing process in an economic view, which analyzes dependence and constraints relations in grid interaction process. Through the study in the paper, researcher can obtain a novel theory on grid organization.(2) Focusing on the problem of applying and matching resources under large-scale users and computing resources in grid environment, a kind of recommendation-based grid resource matching algorithm is presented. Many existing grid resource matching and scheduling algorithms need search and compare every grid computing resource node without considering features of grid resources and users'behaviors, while recommendation system as widely used means in e-commerce could solve all of these two problems well. To utilize recommendation mechanism could pretreat information of users and resources by translating features of grid resources to eigenvectors of items in recommendation system and setting up a satisfaction grade system considering history records with features described in resources applying process that reflected users'behaviors through the frequency users computed in resource nodes. Then, this paper improved SVD-based (Singular value decomposition) and Item-based collaborative filtering algorithm that can give users recommendation resource sets by computing the best approximate resource features to users'behavior features matrix. Especially, the grid resource matching algorithm could mine latent features from given data; efficiently overcome the extreme sparsity of user satisfaction grade data and make use of feedback information from resources scheduling. The paper will solve a mass of resources matching problem in a novel way from new view.(3) Under service oriented grid environment, there are a great amount of various resources described in many different ways. And the traditional methods of grid resource retrieval do not fit the more and more complex grid environment any longer, so it is important to develop a reasonable and effective method to suit future grid system. This paper proposes a novel resource retrieval method which is based on user behavior feedback. And this method analyzes user behavior from history records with relevance feedback technique and establishes a correlation model between grid resources through resource access path, access frequency, access time and retrieval results. This method makes grid resources retrieval results closer to user's subjective perception and reduce interference from complicated and ambiguous semantic environment in heterogeneous grid environment. This grid resource retrieval method achieves a better retrieval precision from simulation experiment from the new view.(4) Since relevance feedback can analyze user behavior to acquire relationship between resources in grid, this paper proposes a novel grid resource retrieval scheme based on association rule hypergraph partitioning algorithm which analyzes user behavior from history records. From those the scheme set up a behavior related retrieval process to meet personality query for the numerous resource in grid. Based on grid resource retrieval transaction database created from past retrieval, it discovers relationship between resources to generate a resource association rule hypergraph. Then the scheme partitions the hypergraph by association rule hypergraph partitioning algorithm that clusters resources based user behavior and semantic. It make grid resources retrieve results closer to user's subjective perception and reduce interference from complicated and ambiguous semantic environment in heterogeneous grid environment. This kind of resource retrieval way base on historic information learning of user retrievals and creates a related resource retrieval translation database, so it has the advantage of low implementation cost, low complexity, and supports for semantic search. This scheme provides a brand new way for resource retrieval and management.(5) Reasonable grid resource organization and discovery are key research issues in grid system, which should affect resource sharing, collaboration efficiency and system optimization in grid. However, existing grid resource organization and discovery approaches do not pay more attention on the interaction and collaboration between members of grid system so that those approaches lead to limitation in system efficiency and adaptability and restriction on development of grid. Therefore, this paper presents an interaction based method to construct grid virtual organization and the corresponding resource discovery method. The method uses information bottleneck principle to analyze and compress the interaction records of the grid nodes and characteristics of resources and services which could mine behavior of nodes and relationship between them. Through this method, we make nodes form a virtual organization those are relativity and similarity in their behaviors and optimize the grid system. Then, according to interaction records of nodes in a virtual organization, we find the association rules within these nodes to speed up resource discovery process and improve efficiency. The method in this paper orients interaction behavior characteristics of members in grid system and is better applicable and expansible. At the same time, it caters for self-adapting and self-organizing of grid. In the end, we prove the two methods by simulation experiments.